Biological Journal of the Linnean Society, 2015, , –. With 4 figures.

Genetic differentiation of populations of the threatened saproxylic beetle Rosalia longicorn, Rosalia alpina (Coleoptera: Cerambycidae) in Central and South-east Europe

LUKAS DRAG1,2*, DAVID HAUCK2,SANDOR BERCES 3, JAKUB MICHALCEWICZ4, 5 6 1,2 LUCIJA SERIC JELASKA , SANDRA AURENHAMMER and LUKAS CIZEK

1Faculty of Science, University of South Bohemia, Branisovska 31, 37005, Ceske Budejovice, Czech Republic 2Biology Centre CAS, v. v. i., Institute of Entomology, Branisovska 31, 37005, Ceske Budejovice, Czech Republic 3Duna-Ipoly National Park Directorate, Kolto utca 21, 1121, Budapest, Hungary 4Institute of Forest Ecosystem Protection, Faculty of Forestry, University of Agriculture, Al. 29 Listopada 46, 31-425, Krakow, Poland 5Faculty of Science, University of Zagreb, Rooseveltov trg 6, 10000, Zagreb, Croatia 6Institute for Animal Ecology and Landscape Planning, Bergmanngasse 22, 8010, Graz, Austria

Received 21 April 2015; revised 16 June 2015; accepted for publication 16 June 2015

Knowledge of patterns of genetic diversity in populations of threatened species is vital for their effective conservation. Rosalia longicorn (Rosalia alpina) is an endangered and strictly protected beetle. Despite a marked decline in part of its range, the beetle has recently expanded to the lowlands of Central Europe. To facilitate a better understanding of the species’ biology, recent expansion and more effective conservation measures, we investigated patterns of genetic structure among 32 populations across Central and South-east Europe. Eight microsatellite loci and a partial mitochondrial gene (cytochrome c oxidase subunit I) were used as markers. Both markers showed a significant decline in genetic diversity with latitude, suggesting a glacial refugium in north- western Greece. The cluster analysis of the nuclear marker indicated the existence of two genetically distinct lineages meeting near the border between the Western and Eastern Carpathians. By contrast, one widespread mtDNA haplotype was dominant in most populations, leading to the assumption that a rapid expansion of a single lineage occurred across the study area. The genetic differentiation among populations from the north- western part of the study area was, however, surprisingly low. They lacked any substructure and isolation-by- distance on a scale of up to 600 km. This result suggests a strong dispersal capacity of the species, as well as a lack of migration barriers throughout the study area. That the lowland populations are closely related to those from the nearby mountains indicates repeated colonization of the lowlands. Our results further suggest that R. alpina mostly lives in large, open populations. Large-scale conservation measures need to be applied to allow for its continued existence. © 2015 The Linnean Society of London, Biological Journal of the Linnean Society, 2015, 00, 000–000.

ADDITIONAL KEYWORDS: beech – conservation – Natura 2000 – phylogeography – post-glacial recolo- nization – xylophagous insect.

INTRODUCTION development of wide-scale conservation strategies and the use of management actions according to the An understanding of the patterns of genetic diversity current needs (Avise et al., 1987; Moritz, 1994). Pop- in populations of threatened species enables the ulation genetic structure is determined by genetic isolation, which is governed by the forces of genetic *Corresponding author. E-mail: [email protected] drift, natural selection, and gene flow (Slatkin,

© 2015 The Linnean Society of London, Biological Journal of the Linnean Society, 2015, , – 1 2 L. DRAG ET AL.

1987). Furthermore, restrictions in the landscape known to occur in Western (Picard, 1929) and South- permeability can reduce gene flow between habitat eastern (Serafim & Maincan, 2008) Europe. Recently, patches and decrease the effective population size. they have also been repeatedly reported in Central Such populations are then more prone to the effects Europe, mostly from the floodplains of the Danube of genetic drift that decrease genetic diversity and and its tributaries (Jendek & Jendek, 2006; Cizek inhibit adaptability to a changing environment et al., 2009; Hovorka, 2011). As a result of a lack of (Frankham, Ballou & Briscoe, 2002). Simple conser- earlier records, it has been proposed that the beetle vation of species that neglects to consider its popula- spread to the lowlands of Central Europe only tion structure may thus fail to address isolated and recently (Cizek et al., 2009). Such a sudden expan- genetically eroded populations or, by contrast, reveal sion into the previously unexploited habitat accompa- the reservoirs of genetic diversity such as glacial nied by a switch in the host plant may indicate the refugia. Consequently, a broad range of molecular existence of distinct ecotypes of the species, associ- techniques have been used to understand the nature ated with upland and lowland forests or beech and of population structure, and they become an impor- other hosts. tant tool in many studies of threatened and protected Two scenarios for the colonization of Central species (Sunnucks, 2000; Morin, Luikart & Wayne, European lowlands were proposed, including down- 2004; Behura, 2006). slope colonization by nearby upland populations Saproxylic insects are a diverse group with high and colonization by a lowland population from else- ecological and economical importance (Grove, 2002). where (Cizek et al., 2009). A shift of upland popula- Genetic structure has occasionally been studied in tions to lowlands would either require a change in the important pest species (Horn et al., 2006, 2009; the species’ host and/or habitat preference or in the Salle et al., 2007; Carter, Smith & Harrison, 2010). quality of the newly-colonized habitat. Colonization The saproxylic guild, however, also contains a num- by a lowland population originating outside the ber of threatened species, and some of them serve as region would suggest the existence of a lowland lin- models for ecology, conservation biology, and/or as eage, adapted to different habitats and hosts. The umbrella species (Ranius, 2002; Buse, Schroder & host-associated population structure, in which popu- Assmann, 2007). Despite that, very little is known lations exploiting different resources are genetically about their population genetics, and this may com- distinct (Stireman, Nason & Heard, 2005; Ferrari promise conservation efforts; but see also Cox et al. et al., 2012), is well documented in phytophagous (2013); Oleksa et al. (2013); Solano et al. (2013); insects and demonstrates how the environment can Drag & Cizek (2014); Oleksa et al. (2015). impact gene flow, even in the absence of physical The Rosalia longicorn (Rosalia alpina; Linnaeus, barriers (Feder et al., 1994; Via, Bouck & Skillman, 1758) is an endangered and strictly protected saprox- 2000). ylic beetle. It is listed as a priority species under the In the present study, we analyzed parts of nuclear European Union (EU) Habitats Directive, which (microsatellites) and mitochondrial (cytochrome c oxi- makes it an icon of invertebrate conservation in Eur- dase subunit I; COI) DNA of > 30 populations of ope. Its distribution range covers most of Europe; R. alpina from Central and South-east Europe aim- from the Pyrenees, the Alps, and the Carpathians, to ing to identify the patterns of their genetic diversity Crimea, the Caucasus, and the Urals (Sama, 2002). and phylogeographical structure. A broad range of In the south, the beetle reaches Corsica, Sicily, the statistical methods were applied to test hypothe- Greece, and the Turkish province of Hatay. In the ses about presumed refugia and species history. By north, the species has experienced substantial comparing the genetic structure of populations from retreat because it has disappeared from Scandinavia, lowlands and nearby mountains, we attempted to most of Germany, Poland, and Czech Republic reveal the relationship between populations originat- (Slama, 1998; Lindhe, Jeppsson & Ehnstrom,€ 2011; ing from different habitats. Information on patterns Michalcewicz & Ciach, 2015). Despite a notable of genetic diversity in R. alpina populations may decline, the distribution of the species is rather con- increase our knowledge of the species’ biology and tinuous in two large mountain systems: in the Alps facilitate more effective conservation. and the Carpathians, as well as on the Balkan Peninsula (Slama, 1998; Gepp, 2002; Duelli & Wer- melinger, 2005). MATERIAL AND METHODS Although generally considered a montane species associated with European beech Fagus sylvatica L. SAMPLE COLLECTION AND DNA EXTRACTION (Heyrovsky, 1955; Slama, 1998), R. alpina also The material analyzed included beetles from 33 local- inhabits lowlands and utilizes a wide range of broad- ities in Central and South-east Europe, thus covering leaved trees. The lowland populations were for long a significant part of the species’ distribution in

© 2015 The Linnean Society of London, Biological Journal of the Linnean Society, 2015, , – POPULATION GENETIC STRUCTURE OF ROSALIA ALPINA 3

A

B

Figure 1. A, localization and the genetic structure of 32 populations (full names are listed in Table 1) of the Rosalia longicorn (Rosalia alpina) based on eight microsatellite loci. Each pie chart represents a proportion of membership of individuals from a given population in each of the two clusters indicated by the Bayesian clustering analysis (STRUC- TURE). Grey surface represents the area above 800 m a.s.l. Upper barplot illustrates the division of each individual into two colours, reflecting the estimated assignment into two clusters. Solid black lines define the boundaries between the same populations as in the map. B, distribution range of R. alpina (light green colour) and the study area (black rectangle).

Europe (Fig. 1B). Samples originated from both the MICROSATELLITES – uplands (altitudinal range 300 1050 m a.s.l.; 27 pop- Amplification – ulations) and the lowlands (0 300 m a.s.l.; seven In total, 700 individuals were genotyped for nine populations). The upland sites were dominated by polymorphic microsatellite loci previously described beech (Fagus spp.) forests, and the beetles were col- by Drag, Zima & Cizek (2013). Because of an insuffi- lected on beech wood. In the lowlands, the beetles cient sample size, one population was excluded from were always associated with other tree species, the microsatellite’s dataset, thus leaving a total of 32 including elms (Ulmus spp.), maples (Acer spp.), and populations and 695 individuals. The locus RA_29 ash (Fraxinus spp.). Details on the localities, host repeatedly failed to amplify in many individuals of plants, and number of individuals analyzed are pro- some populations. We thus decided to exclude it from vided in Table 1. A part of a middle leg from all dis- further analyses. Hence, all of the reported results in covered specimens (dead or alive) was taken and the presents study are based on eight loci. Poly- stored in vials containing 96% ethanol for molecular merase chain reaction (PCR) products were analyzed analyses. Genomic DNA was extracted from the sam- with an automated sequencer ABI 3730XL (Applied pled tissue using the Genomic DNA Mini Kit Tissue Biosystems) by a commercial company (Macrogen (Geneaid) in accordance with the manufacturer’s Inc.). Allelic patterns were scored using GENEMAP- instructions. PER, version 3.7 (Applied Biosystems).

© 2015 The Linnean Society of London, Biological Journal of the Linnean Society, 2015, , – 4 L. DRAG ET AL.

Table 1. Sampled populations of the Rosalia longicorn (Rosalia alpina) with information about their habitat, host tree, GPS coordinates, and the number of individuals analyzed for eight microsatellite loci and cytochrome c oxidase I (COI)

Number of individuals analyzed

Country Locality Code Habitat Host tree Latitude Longitude COI Microsatellites

Austria Kalkalpen AT1 U FS 47.805571 13.950015 5 20 Austria Wienerwald AT2 U FS 48.010122 16.199364 5 20 Bulgaria Stara planina II BG1 U FS 42.784421 23.790154 5 20 Bulgaria Strandja BG2 U FO 42.08869 27.750227 5 20 Bulgaria Ropotamo BG3 L AC, FA 42.297936 27.724085 5 20 Czech Dyje floodplain CZ1 L AC, UL 48.718113 16.892834 5 18 Republic Czech Bezdez CZ2 U FS 50.539185 14.720318 5 29 Republic Czech Bile Karpaty CZ3 U FS 49.032422 18.025136 5 18 Republic Greece Olymp GR1 U FS 40.108456 22.460764 5 20 Greece Pindos GR2 U FS 39.959741 20.906086 5 16 Greece Vermio GR3 U FS 40.589685 22.042605 5 7 Greece Rodopi GR4 U FS 40.921661 24.189622 5 20 Greece Evros GR5 U FS 41.109798 25.962128 5 9 Croatia Lonsko Polje HR1 L FA 45.196004 17.128523 5 18 Croatia Mt. Medvednica HR2 U FS 45.88291 15.953268 5 17 Hungary Kab-Hegy HU1 U FS 47.04975 17.655917 5 41 Hungary Bakony HU2 U FS 47.215248 17.666995 5 21 Hungary Borzsony HU3 U FS 47.917322 18.977368 5 20 Hungary Pilis HU4 U FS 47.721766 18.960977 5 19 Hungary Mecsek HU5 U FS 46.21439 18.355765 5 32 Poland Beskid Nisky PL1 U FS 49.446583 21.184611 5 20 Romania Comana forest RO1 L AC, T, F 44.15751 26.100216 5 18 Romania Apuseni RO2 U FS 46.461244 23.374803 5 20 Slovakia Bratislava SK1 U FS 48.203872 17.0931 5 30 Slovakia Dunajske Luhy SK2 L AC, AP, 48.080593 17.179173 5 36 UL Slovakia Male Karpaty SK3 U FS 48.327994 17.220819 5 45 Slovakia Devinska Kobyla SK4 L FE 48.186665 17.005319 5 27 Slovakia Kovacov SK5 L Q 47.827912 18.75877 5 22 Slovakia Strazovske vrchy SK6 U FS 48.768333 18.427567 5 8 Slovakia Muranska SK7 U FS 48.760975 20.050352 4 30 planina Slovakia Vihorlat SK8 U FS 48.886894 22.242261 5 26 Serbia Stara planina SR1 U FS 44.173172 22.123664 5 0 Ukraine Crimea UKR1 U FO 44.74788 34.333431 5 8

U, uplands; L, lowlands; FS, Fagus sylvatica; FO, Fagus orientalis; AC, Acer campestre; AP, Acer pseudoplatanus; UL, Ulmus laevis; FA, Fraxinus angustifolia; FE, Fraxinus excelsior;F,Fraxinus sp.; T, Tilia sp.; Q, Quercus sp.

Loci characteristics and genetic diversity 4.1.3 (Raymond & Rousset, 1995; Rousset, 2008) Frequencies of null alleles were estimated using using the default parameters. FREENA (Chapuis & Estoup, 2007). The linkage dis- For each population, we calculated the number of equilibrium between all pairs of loci, as well as the alleles (NA), observed (HO) and expected (HE) Hardy–Weinberg equilibrium (HWE) across loci and heterozygosity, and the number of private alleles populations, was tested with GENEPOP, version (NPA) using GENALEX, version 6.5 (Peakall &

© 2015 The Linnean Society of London, Biological Journal of the Linnean Society, 2015, , – POPULATION GENETIC STRUCTURE OF ROSALIA ALPINA 5

Smouse, 2006, 2012). Allelic richness (AR) corrected FST/(1 – FST). The significance of correlation was for sample size variability was computed in FSTAT, tested by a Mantel test (Mantel, 1967) using the version 2.9.3.2 (Goudet, 2002). Population specific IBDWS (Jensen, Bohonak & Kelley, 2005) with coefficient of inbreeding (FIS) was calculated and 10 000 permutations. IBDWS calculates the slope tested using ARLEQUIN, version 3.5.1.3 (Excoffier & and intercept of the isolation-by-distance (IBD) rela- Lischer, 2010) with 10 000 permutations. Two tionship using reduced major axis regression, which indices (HE and AR), which represented the genetic is more appropriate than standard linear regression diversity of populations, were used in a linear regres- (Bohonak, 2002). IBD patterns were calculated for sion as a function of latitude and longitude. all populations as well as within the north-western (NW) and south-eastern (SE) lineages gained from Population structure STRUCTURE. The UKR1 population was omitted The population structure of our dataset was assessed from the IBD analyses as a result of the low number using STRUCTURE, version 2.3.4 (Pritchard, Ste- of samples available for this highly remote and iso- phens & Donnelly, 2000), assuming an admixture lated population. model (individuals may have mixed ancestry) and The gene flow among populations was estimated as correlated allele frequencies (closely-related popula- pairwise FST values calculated in ARLEQUIN, ver- tions might have correlated allele frequencies). We sion 3.5.1.3 (Excoffier & Lischer, 2010). The signifi- ran STRUCTURE for values of K ranging from 1 to cance of the derived genetic distances was tested by 10 with 100 000 burn-in and 1 000 000 Markov 10 000 permutations. To avoid any potential bias by chain Monte Carlo (MCMC) steps for 10 replicates null alleles, we also calculated the pairwise FST val- for each K. The best K value was chosen according to ues using a correction for null alleles method imple- Evanno, Regnaut & Goudet (2005), using the mented in FREENA (Chapuis & Estoup, 2007). STRUCTURE HARVESTER (Earl & vonHoldt, 2012). The results obtained for a given K were post- processed in CLUMPP, version 1.1.2 (Jakobsson & COI Rosenberg, 2007) and used to generate pie charts, Amplification illustrating the geographical structure of each popu- A partial fragment of the mitochondrial gene for COI lation. Barplots were visualized in DISTRUCT, ver- (approximately 766 bp in length) was amplified and sion 1.1 (Rosenberg, 2004). sequenced for 164 individuals from the 33 popula- tions. Five individuals represented each population, Genetic differentiation with the exception of SK7, which was composed of All populations were assigned into two groups four individuals. We used universal forward and according to their habitat (‘lowland’ and ‘upland’) reverse primers: C1-J-2183 (alias Jerry) (50-CAA (Table 1). Genetic differentiation among populations CAT TTA TTT TGA TTT TTT GG-30) and TL2-N- within and between the two groups was tested using 3014 (alias Pat) (50-TTC AAT GCA CTT ATT CTG analysis of molecular variance (AMOVA) in ARLE- CCA TAT TA-30) (Simon et al., 1994). PCR cycling QUIN, version 3.5.1.3 (Excoffier & Lischer, 2010) parameters included a denaturation step at 95 °C for with 10 000 permutations. 5 min and 40 cycles at: 92 °C for 30 s, 50 °C for 30 s, To construct the phylogenetic tree of populations, and 72 °C for 1 min 30 s. The cycling concluded with we performed the evolutionary analysis of allele fre- a final elongation step at 72 °C for 10 min. Sequenc- quencies using a neighbour-joining (NJ) method in ing was performed by a commercial company (Macro- POPTREE2 (Takezaki, Nei & Tamura, 2010) with gen Inc.). 10 000 bootstrap replicates. As genetic measures, we used DA distance values (Nei & Chesser, 1983). Fur- Genetic diversity and differentiation thermore, alternate indices based on DST distance or The sequences from each individual were edited and FST produced congruent results (data not shown). aligned (MUSCLE, default settings) in GENEIOUS, Finally, we adjusted the constructed NJ tree in version 6.1.6 (Biomatters). To minimize the probabil- MEGA, version 6 (Tamura et al., 2013). ity of the occurrence of nuclear mitochondrial pseudo- genes in our data, we carefully checked all sequences Gene flow for the presence of double peaks, indels, frameshifts, The pairwise geographical distances between all pop- and stop codons, as suggested by Song et al. (2008). ulations were computed from the list of coordinates Although such measures cannot completely rule out using the GEOGRAPHIC DISTANCE MATRIX their presence, they can minimize any possibility and GENERATOR, version 1.2.3 (Ersts, 2015). Their also prevent overestimation of the genetic diversity logarithmic values (log) were plotted against indices and misinterpretation of the phylogeography the linearized form of pairwise genetic distances (Haran et al., 2015). All unique sequences were

© 2015 The Linnean Society of London, Biological Journal of the Linnean Society, 2015, , – 6 L. DRAG ET AL. submitted to GenBank (accession number: KT351997– Population structure KT352027). For each population, standard genetic According to DK (Evanno et al., 2005), the best num- indices such as the number of haplotypes (H), the ber of clusters was identified as two (K = 2) in haplotype (h) and nucleotide (p) diversities, and the STRUCTURE (see Supporting information, Fig. S1). number of polymorphic sites (P) were computed using It divided populations from the study area into the DNASP, version 5.10 (Librado & Rozas, 2009). Two NW and SE lineages (Fig. 1A). The boundary indices (h and p) representing the genetic diversity of between the two lineages was not well defined, and populations (similar to those used for microsatellites) nearby populations represented a mixture of these were used in a linear regression. A haplotype network two clusters. We were unable to detect further sub- was produced using the statistical parsimony method structure when K > 2, except by the clear separation (95% connection limit; Templeton, Crandall & Sing, of three Greek populations (GR1, GR2, GR3) for 1992), implemented in TCS, version 1.21 (Clement, K ≥ 4 (see Supporting information, Fig. S1). Posada & Crandall, 2000). The genetic differentiation among populations inhabiting different habitats (‘low- Genetic differentiation land’ and ‘upland’) was estimated using AMOVA in AMOVA showed no genetic structure between the ARLEQUIN, version 3.5.1.3 (Excoffier & Lischer, ‘lowland’ and ‘upland’ habitat groups (0.6%,

2010) with 10 000 permutations. FCT = 0.0055, P = 0.633), and little variation among populations within these groups (12.1%,

FSC = 0.1205, P < 0.0001). Most of the variation in the microsatellites might be explained by variation RESULTS within populations (88.4%, FST = 0.1156, P < 0.0001). MICROSATELLITES The NJ tree, based on DA distances among popula- Loci characteristics and genetic diversity tions was congruent with the results gained from In total, 695 individuals from 32 populations were STRUCTURE. Although the bootstrap supports of genotyped at eight microsatellite loci. All analyzed some branches were rather weak, there was an loci were polymorphic, with the number of alleles per apparent division between the populations inhabiting locus ranging from five to sixteen (mean 8.25). With NW and SE parts of the study area (Fig. 2). the exception of RA_37 paired with: RA_11 (HU5; P = 0.00404), RA_08 (RO2; P = 0.00359), and RA_13 Gene flow (SK8; P = 0.00333), no linkage disequilibrium was Genetic and geographical distances were correlated found between all pairs of loci for each population among all populations (r = 0.632, Mantel test: after Bonferroni correction for multiple tests. The P < 0.001), as well as among populations belonging mean estimated frequency of null alleles per locus to the SE lineage (r = 0.406, Mantel test: P = 0.016) across all populations was never higher than 15%. (Fig. 3). Within the NW lineage, however, we found However, in some populations and some loci, the no relationship between geographical and genetic presence of null alleles exceeded such probability distances (r = 0.193, Mantel test: P = 0.092), despite (see Supporting information, Table S1). the fact that the longest geographical distance Populations with the highest genetic diversity between two populations was more than 620 km. were found in Greece, with the mean number of alle- FST values were significantly different (P < 0.05) in les per locus ranging from 3.1 to 4.8. The lowest 88.3% of all pairwise comparisons. They showed a diversity was observed in UKR1 (1.8), CZ2 (2.3), and rather variable amount of genetic differentiation rang- CZ1 (2.3) (Table 2). We found the same pattern for ing from values close to zero (many pairs of populations allelic richness (AR) when we considered the varia- in the northern area) to 0.44 (between GR3 and UKR1) tion in population sizes. The mean observed and (see Supporting information, Table S2). We achieved expected heterozygosity of all loci ranged from 0.172 similar results after the ENA (= excluding null alleles) to 0.602, and from 0.167 to 0.663, respectively. We correction (Chapuis & Estoup, 2007), with the highest found deviations from HWE for 17 populations (after value of FST = 0.43 between GR3 and UKR1. Bonferroni correction), with the maximum of three deviating loci per population. Private alleles were recorded for twelve populations, with the frequencies COI ranging from 0.011 to 0.094. We found a significant Genetic diversity and differentiation decrease in genetic diversity with increasing latitude We identified 31 different haplotypes based on the for both HE (R = 0.397, P < 0.05) and AR (R = 0.659, 766-bp long fragment of the mitochondrial gene COI. P < 0.05). On the other hand, we did not find any The overall haplotype diversity (h) was 0.541 and evidence for a change in HE or AR on the longitudi- the nucleotide diversity (p) was 0.0016 (Table 2). nal gradient. The highest diversity values (both haplotype and

© 2015 The Linnean Society of London, Biological Journal of the Linnean Society, 2015, , – POPULATION GENETIC STRUCTURE OF ROSALIA ALPINA 7

Table 2. Genetic diversities of the sampled populations of the Rosalia longicorn (Rosalia alpina) derived from eight microsatellite loci and cytochrome c oxidase I (COI)

Microsatellites COI

Population NA AR HO HE NPA FIS Hh p P

AT1† 2.6 2.38 0.378 0.458 0 0.194** 1 0 0 0 AT2 2.8 2.57 0.404 0.473 0 0.164* 3 0.7 0.00104 2 BG1† 3.0 2.50 0.320 0.450 0 0.307*** 2 0.4 0.00104 2 BG2 † 3.3 2.64 0.277 0.408 0 0.322*** 2 0.4 0.00052 1 BG3† 3.4 2.73 0.286 0.434 2 0.354*** 1 0 0 0 CZ1 2.3 2.11 0.333 0.362 0 0.108 1 0 0 0 CZ2 2.3 2.00 0.297 0.318 0 0.083 1 0 0 0 CZ3 2.9 2.55 0.395 0.454 0 0.140* 1 0 0 0 GR1† 4.8 3.84 0.456 0.585 2 0.245*** 4 0.9 0.00468 8 GR2† 4.6 4.06 0.602 0.663 1 0.124* 5 1 0.00649 10 GR3 3.6 3.63 0.500 0.560 0 0.182* 4 0.9 0.0026 4 GR4† 3.6 2.83 0.338 0.440 1 0.256*** 2 0.4 0.00208 4 GR5 3.1 3.01 0.375 0.511 0 0.320** 2 0.4 0.00052 1 HR1 2.6 2.47 0.465 0.464 0 0.025 1 0 0 0 HR2 2.9 2.53 0.347 0.429 0 0.210** 2 0.4 0.00052 1 HU1† 3.0 2.48 0.442 0.445 0 0.019 2 0.4 0.00052 1 HU2† 3.0 2.57 0.369 0.445 1 0.194** 1 0 0 0 HU3† 2.9 2.62 0.394 0.456 1 0.162* 2 0.4 0.00052 1 HU4† 3.0 2.66 0.349 0.463 0 0.272*** 2 0.4 0.00052 1 HU5 2.9 2.49 0.414 0.464 1 0.123* 3 0.7 0.00104 2 PL1† 3.0 2.72 0.438 0.491 0 0.135* 2 0.6 0.00078 1 RO1† 3.4 2.93 0.375 0.545 0 0.314*** 2 0.6 0.00078 1 RO2 3.0 2.57 0.413 0.454 1 0.116 3 0.8 0.00156 2 SK1† 3.0 2.41 0.375 0.443 0 0.170** 2 0 0 1 SK2† 2.8 2.23 0.345 0.414 1 0.161** 2 0.4 0.00052 1 SK3 3.3 2.43 0.372 0.434 1 0.154** 1 0 0 0 SK4† 3.1 2.64 0.435 0.491 0 0.123* 1 0 0 0 SK5 3.0 2.54 0.392 0.432 0 0.115 2 0.4 0.00052 1 SK6 2.6 2.58 0.453 0.442 0 0.042 2 0.4 0.00052 1 SK7† 3.5 2.77 0.363 0.478 1 0.258*** 2 0.5 0.00065 1 SK8 3.3 2.61 0.428 0.450 2 0.069 3 0.8 0.0013 2 SR1 2 0.4 0.00052 1 UKR1 1.8 1.70 0.172 0.167 0 0.038 2 0.4 0.00052 1

NA, mean number of alleles across the eight loci; AR, mean allelic richness per locus, based on a minimum of seven indi- viduals; HO, observed heterozygosities; HE, expected heterozygosities; NPA, number of private alleles; FIS, genetic simi- larity of individuals within the population (*P < 0.05, **P < 0.01, ***P < 0.001); H, number of haplotypes; h, haplotype diversity; p, nucleotide diversity; P, number of polymorphic sites. †Significant deviation from Hardy–Weinberg equilibrium. nucleotide) were found in the three Greek popula- frequently present, especially in the Carpathian tions (GR1, GR2, GR3). As for microsatellites, we Mountains (H3; 10% of all individuals), and many observed a significant increase in haplotype (R = rare haplotypes that were separated by one or two 0.42, P < 0.05) and nucleotide (R = 0.621, P < 0.05) mutations from the H1 and often unique for a given diversities with decreasing latitude, although no evi- population (Fig. 4). Beside this star-like pattern, all dence for change in either parameters along the lon- haplotypes of the three Greek populations (GR1, gitudinal gradient. GR2, and GR3) formed a separate and more diverse The haplotype network based on the statistical network, where the most common haplotype (H1) parsimony analysis showed one dominating haplo- was missing. Nevertheless, even the most remote type widely present in many of the studied popula- Greek haplotype was not separated by more than tions (H1; 67% of all individuals), one haplotype seven mutations from the central haplotype.

© 2015 The Linnean Society of London, Biological Journal of the Linnean Society, 2015, , – 8 L. DRAG ET AL.

Figure 3. Isolation-by-distance (IBD) analysis of Rosalia longicorn (Rosalia alpina) species. Genetic differentiation

(estimated as FST/1 – FST based on the eight microsatel- lite loci) was plotted against logarithm of geographical distances (km). IBD was constructed across 31 popula- tions (r = 0.632, Mantel test: P < 0.001), within the north-western lineage (r = 0.193, Mantel test: P = 0.092), and within the south-eastern lineage (r = 0.406, Mantel test: P = 0.016). Population UKR1 was excluded.

distributional range using both nuclear and mito- chondrial genetic markers. Our results investigate patterns of genetic diversity of this EU-wide pro- Figure 2. Unrooted neighbour-joining tree of 32 popula- tected species and provide new information about tions of the Rosalia longicorn (Rosalia alpina) based on the beetle’s population biology.

DA distance values from eight microsatellite loci. Boot- strap values above 40% are shown (10 000 replicates). Phylogeography and population structure Full names of the locations are listed in Table 1. We observed a significant decline in genetic diversity of both markers with latitude. This is most likely a Similar to the microsatellite data, AMOVA showed consequence of the gradual loss of genetic variation no genetic structure between the ‘lowland’ and ‘up- with increasing distance from the glacial refugium. land’ habitat groups (3.1%; FCT = 0.03114, P = 0.116) This form of ‘southern richness and northern purity’ and most of the variation was found within popula- is a common pattern among temperate species tions (60.4%; FST = 0.1156, P < 0.0001). On the other caused by post-glacial colonization (Hewitt, 1999) hand, variation among populations within habitat and frequently found in many taxons (Comps et al., groups was substantially higher than for microsatel- 2001; Gassert et al., 2013; Wielstra et al., 2013; lites (36.5%; FSC = 0.3766, P < 0.0001). Tison et al., 2014). Thus, rather unsurprisingly for a beetle that depends on broadleaved trees, R. alpina is among the species that expanded during the warm periods of the Pleistocene cyclic climate changes. It DISCUSSION means that R. alpina is not technically an ‘alpine’ The present study represents the first analysis of species (sensu Schmitt, 2007). spatial genetic variations of the endangered Rosalia The exceptional high AR and haplotype diversity longicorn beetle, R. alpina. We present an analysis of the southern populations (especially of three of populations from a substantial part of the beetle’s Greek populations GR1, GR2, and GR3) suggests the

© 2015 The Linnean Society of London, Biological Journal of the Linnean Society, 2015, , – POPULATION GENETIC STRUCTURE OF ROSALIA ALPINA 9

A B

Figure 4. A, geographical distribution of the 31 haplotypes among the 33 sampled populations (full names are listed in Table 1) of Rosalia longicorn (Rosalia alpina) based on cytochrome c oxidase subunit I (COI) sequences. Each pie chart represents a proportion of haplotypes from a given population. Same haplotypes were marked by same colour (white parts represent haplotypes unique for a given population). Grey surface represents the area above 800 m a.s.l. B, haplo- type network constructed using the statistical parsimony method (TCS); each haplotype is represented by a circle; the circle colour corresponds to the haplotype distribution map. The circle size is proportional to the haplotype frequency. Small black circles indicate missing haplotypes that are not present in the sample but are necessary to link all observed haplotypes to the network. Blue area highlights all haplotypes found only in three Greek populations (GR1, GR2, GR3). existence of a glacial refugium of R. alpina in the Based on the microsatellites, the cluster analysis high mountains of NW Greece (Pindos, Olymp). Our indicated the existence of two genetically distinct lin- data do not enable us to determine whether the refu- eages. The SE lineage consists of populations from gium was restricted only to the higher mountains of Greece, Bulgaria, Romania, eastern Slovakia, and NW Greece, or whether it also reached the moun- Poland. This lineage probably originates in the afore- tains along the eastern coast of the Adriatic Sea. The mentioned glacial refugium. The NW lineage former scenario of the small glacial refugium is in includes populations from central Austria, Czech line with the mountains of Greece acting as refugia Republic, western Slovakia, Hungary, Croatia, and, for numerous other invertebrate species (Gratton, surprisingly, Ukraine (Crimea). The presence of the Konopinski & Sbordoni, 2008; Dinca et al., 2013; second lineage suggests the existence of another gla- Theissinger et al., 2013). Nevertheless, the latter sce- cial refugium for this species. As has been shown for nario of a larger refugium is supported by the high other organisms (Schmitt, 2009), there is no signifi- level of endemism on the eastern Adriatic coast and cant boundary between the Eastern Alps and the the presence of numerous glacial refugia there, Western Carpathians. For R. alpina, it is rather including probably one of the beetle’s main hosts: the unsurprising because the shortest distance between European beech (Fagus sylvatica) (Magri et al., 2006; the studied populations from two montane systems Magri, 2008; Brus, 2010). (Wienerwald with population AT2 and Lesser

© 2015 The Linnean Society of London, Biological Journal of the Linnean Society, 2015, , – 10 L. DRAG ET AL.

Carpathians with population SK3) is an approxi- differentiation of populations from the NW lineage of mately 50 km and the beetle also inhabits the low- R. alpina was, however, surprisingly low, despite land area between them. Our results suggest the two that the analysis involved populations from sites as lineages met somewhere near the border between far as 620 km apart. the Western and Eastern Carpathians. Such a pat- The genetically poorest population was the one tern in which the Carpathians were probably colo- inhabiting Crimean Peninsula (UKR1). Although its nized from several sources and acted as the contact low AR might be partly resulting from the small zone has already been described for other insects sample size, it requires additional explanation. Dur- (e.g. the meadow spittlebug Philaenus spumarius) ing the last glaciation, the Crimea was mostly cov- (Lis et al., 2014). ered by steppe vegetation (Atanassova, 2005) and the The mitochondrial (mt)DNA haplotype diversity persistence of full forest cover over a significant area also decreases with increasing latitude. This supports is rather unlikely (Cameron, Pokryszko & Horsak, the assumption of a glacial refugium in NW Greece, 2013). This, together with the low AR and no private thus corroborating the aforementioned results of alleles in the R. alpina population, suggests the nuclear DNA analyses. There are, however, some dif- post-glacial recolonization of the Crimea by a limited ferences in the outcomes of the analyses of the two number of immigrants. The second-least diverse pop- markers. The central mtDNA haplotype H1 was wide- ulation was found in the northern Czech Republic spread and dominant in most of the study area. This (CZ2). The population is confined to a very small may suggest that this area was colonized through the area, having been isolated for decades from other rapid expansion of a single lineage. The lineage might known populations by hundreds of kilometres (Drag have originated in NW Greece, despite the H1 haplo- et al., 2011). Although this population is currently type not being detected there. Given the high haplo- relatively large, its low AR might be explained by type diversity in NW Greece, the rare presence of the the fluctuation in population size. H1 haplotype is likely in this region. It is interesting We failed to find any substructure within the NW that all haplotypes found in NW Greece were lineage (NJ tree, cluster analysis). Despite all popu- restricted to this area; they probably did not con- lations within the NW lineage being genetically tribute to the rest of the European gene pool, although rather poor, all of the microsatellite loci were poly- the possibility of their spread along the Adriatic Sea morphic. The populations were nevertheless surpris- cannot be ignored. Furthermore, the occurrence of the ingly uniform and displayed low incidence of private H3 haplotype was confined to the Carpathians and alleles. Thus, the lack of substructure within the lin- their vicinity. It is unclear whether the haplotype is a eage requires another explanation than that of the result of a single mutation in H1 that may have low genetic diversity. This suggests high population occurred during the colonization process, or whether admixture over most of Central Europe and/or con- it represents a separate lineage with its own refugium servatism in genetic structure. located in the Carpathians (as suggested for beech; The high admixture suggests high gene flow of Magri, 2008) or somewhere outside the study area. R. alpina populations within the NW lineage. It Interestingly, the transition between haplotypes H1 might be a result of the strong dispersal capacity of and H3 also occurs near the border between the Wes- the species, the lack of migration barriers, and/or tern and Eastern Carpathians, thus partly corroborat- wood trade related accidental translocations. Genetic ing the results of the nuclear markers. diversities of R. alpina do not resemble other saprox- The distinctions between nuclear and mitochondrial ylic beetles with restricted dispersal ability such as markers are common and may be influenced by pat- flightless longhorn beetle Morimus funereus Mul- terns of, for example, mating, sex-biased dispersal sant, 1862 (Solano et al., 2013) or saproxylic Hermit (Chesser & Baker, 1996; Miller, Haig & Wagner, 2005; beetle Osmoderma barnabita Motschulsky, 1845 Caparroz, Miyaki & Baker, 2009) or different evolu- (Oleksa et al., 2013). They are rather similar to tionary dynamics of the markers (Frankham et al., highly mobile widespread saproxylic pests such as 2002). Despite some discordance in results of both bark beetles (Cognato, Seybold & Sperling, 1999; markers, the main patterns were similar. We may Cognato, Harlin & Fisher, 2003; Avtzis, Arthofer & thus conclude that the populations in NW Greece Stauffer, 2008; Horn et al., 2009) or some rather deserve particular conservation efforts because they mobile butterflies (Zakharov & Hellmann, 2008). are the major genetic diversity reservoir of the species. Furthermore, considering the spatial scale investi-

gated, the FST values based on microsatellites were Low genetic structure of the NW lineage rather low, typical for butterflies with high dispersal The IBD was significant across all populations abilities (Williams, Brawn & Paige, 2003; Zakharov (UKR1 omitted), probably as a result of the higher & Hellmann, 2008; Vandewoestijne & Van Dyck, genetic differentiation of the SE lineage. The genetic 2010). Although the human-related translocations

© 2015 The Linnean Society of London, Biological Journal of the Linnean Society, 2015, , – POPULATION GENETIC STRUCTURE OF ROSALIA ALPINA 11 might have contributed to this pattern, our findings upland populations of R. alpina. Although the low- are in agreement with the results of previous stud- land populations from Western and South-eastern ies, suggesting that the R. alpina is an active, mobile Europe were known for long (Picard, 1929; Serafim species (Drag et al., 2011). Moreover, the species’ & Maincan, 2008), the species most likely appeared current distribution is continuous in many parts of recently in Central European lowlands (Cizek et al., the study area (e.g. Carpathians) and, together with 2009). Indeed, the low AR and presence of the single, its ability to also exploit lowland habitats (see below), otherwise rare mtDNA haplotype H3 in CZ1 (the this indicates a lack of migration barriers within the population referred to by Cizek et al., 2009) imply studied area during most of recent history. the founder effect and the population’s recent estab- With respect to the above, several populations with lishment by a limited number of individuals, proba- presumably restricted gene flow (the distance to the bly from the Carpathians. The recent establishment closest known population > 50 km) were analyzed hypothesis is also supported by further spread of the (CZ2, UKR1, RO1, HU5). They were, however, genet- population (Hovorka, 2011). ically similar to other populations and lacked private Two scenarios for colonization of Central European alleles. This indicates a minimal effect of genetic lowlands including colonization from nearby moun- drift typical for large populations (Frankham et al., tains and colonization by lowland population from 2002). We hypothesize that the effective size of all the south were proposed by Cizek et al. (2009). Our presumably isolated populations was rather large results, based on the comparison of seven lowland during their existence, thus mitigating the effect of and 26 upland populations, ruled out the colonization genetic drift. This would indicate that R. alpina is from the south and thus also militate against the unable to exist in small, isolated populations, possi- existence of a separate lowland lineage. Most low- bly as a result of some intrinsic reasons related to land populations in Central Europe were found the species’ biology or behaviour. Such constraints to within a few kilometres of the nearest upland popu- population survival are common in other groups, lation. This, together with the genetic composition, including butterflies (Kadlec et al., 2010), although suggests independent colonization events from the their existence has not been described for beetles so nearby upland populations. far. This hypothesis would not only explain the sur- The beetle regularly exploits nonbeech hosts in prisingly low genetic differentiation of R. alpina the mountains (Michalcewicz & Ciach, 2012; Michal- within Central Europe, but also might explain why a cewicz, Bodziarczyk & Ciach, 2013). This common mobile species with the ability to exploit a broad ability to use nonbeech hosts most likely features in range of habitats has disappeared from substantial individuals of most, if not all, populations. Thus, no part of its range (see above). shift in host preference as suggested by Cizek et al. The above hypothesis, as well as the high mobility (2009) was needed to allow for the observed of the species inferred from genetic and mark–recap- R. alpina expansion to lowlands. Although the cli- ture (Drag et al., 2011) data, imply that conservation mate change might have some effect (Muller€ et al., measures applied on a large scale are likely to bene- 2015), it is more likely that the expansion of fit the beetle’s conservation more than local mea- R. alpina to the lowlands has been facilitated by sures (Bosso et al., 2013; Fahrig, 2013). More changes in the colonized habitat. Cucujus cinnabari- specifically, creating small patches of suitable habi- nus (Scopoli, 1763) is another example of an endan- tats or corridors between habitat patches only sev- gered saproxylic beetle species that originally eral kilometres distant is less important than inhabited mid and higher altitudes, and has creating matrix of habitats consisting of larger expanded to the lowlands of Central Europe in the patches of suitable habitat within the beetles reach, last few decades (Horak, Chumanova & Hil- most likely up to 5–15 km. It is also important to szczanski, 2012). Although the expansion of C. note that management measures focusing on micro- cinnabarinus has been attributed to an increase of habitat creation on individual trees, such as pollard- the amount of poplar dead-wood, the expansion of ing and shredding (Russo, Cistrone & Garonna, R. alpina to lowlands might be attributed to 2011; Castro et al., 2012; Sebek et al., 2013), are increased availability of deadwood of shade tolerant likely to be more effective than the often recom- trees such as ash (Fraxinus spp.) and maple (Acer mended stand-focused measures such increase of spp.) (Marigo et al., 2000; Garbarino et al., 2014). rotation age or push for changes in tree species com- Although the recent spread of the two species to low- position in large areas. lands of Central Europe is certainly beneficial for them, it may indicate changes in the tree species Uplands versus lowlands composition and spatial structure of lowland forests. Based on both microsatellites and mtDNA, no Such changes have indeed been described as a tran- genetic differences were found between lowland and sition from thermophilous, open, mainly oak

© 2015 The Linnean Society of London, Biological Journal of the Linnean Society, 2015, , – 12 L. DRAG ET AL. woodlands to closed canopy forests with more mesic beetle – a case study for saproxylic insect conservation. Bio- conditions (Hedl, Kopecky & Komarek, 2010; Miklın logical Conservation 137: 372–381. & Cızek, 2014). Some saproxylic species certainly Cameron RAD, Pokryszko BM, Horsak H. 2013. Forest benefited from the change but this has led to a snail faunas from Crimea (Ukraine), an isolated and incom- decrease in the habitat available for the substan- plete Pleistocene refugium. Biological Journal of the Lin- tially more diverse and endangered lowland fauna nean Society 109: 424–433. (Seibold et al., 2014). Caparroz R, Miyaki CY, Baker AJ. 2009. Contrasting phy- logeographic patterns in mitochondrial DNA and microsatellites: evidence of female philopatry and malebi- ased gene flow among regional populations of the blueand- ACKNOWLEDGEMENTS yellow macaw (Psittaciformes: Ara ararauna) in Brazil. The Auk 126: 359–370. We would like to thank C. Komposch, J. Borucky, M. Carter M, Smith M, Harrison R. 2010. Genetic analyses of Platek, P. Butterill, D. Lestina, O. Konvicka, and R. the Asian longhorned beetle (Coleoptera, Cerambycidae, Godunko for assistance with obtaining samples; J. Anoplophora glabripennis), in North America, Europe and Zima for assistance with molecular analyses; A. Hood Asia. Biological Invasions 12: 1165–1182. for language correction of the submitted manuscript; Castro A, Martınez de Murguıa L, Fernandez J, Casis and A. Oleksa and three anonymous reviewers for A, Molino F. 2012. Size and quality of wood used by Ros- their valuable comments on the manuscript. The alia alpina (Linnaeus, 1758) (Coleoptera: Cerambycidae) in study and its authors were supported by Czech beech woodlands of Gipuzkoa (northern Spain). Munibe Science Foundation (P504/12/1952), the Grant Ciencias Naturales 60: 77–100. Agency of the University of South Bohemia (04-168/ Chapuis MP, Estoup A. 2007. Microsatellite null alleles 2013/P), and the Project Biodiversity of Forest and estimation of population differentiation. Molecular Ecosystems (CZ.1.07/2.3.00/20.0064), co-financed by Biology and Evolution 24: 621–631. the European Social Fund and the state budget of Chesser RK, Baker RJ. 1996. Effective sizes and dynamics the Czech Republic. of uniparentally and biparentally inherited genes. Genetics 144: 1225–1235. Cizek L, SchlaghamerskyJ,Bo ruck y J, Hauck D, Helesic J. 2009. Range expansion of an endangered beetle: REFERENCES alpine longhorn Rosalia alpina (Coleoptera: Cerambycidae) Atanassova A. 2005. Palaeoecological setting of the western spreads to the lowlands of Central Europe. Entomologica – Black Sea area during the last 15 000 years. The Holocene Fennica 20: 200 206. 15: 576–584. Clement M, Posada D, Crandall K. 2000. TCS: a computer Avise JC, Arnold J, Ball RM, Bermingham E, Lamb T, program to estimate gene genealogies. Molecular Ecology 9: – Neigel JE, Reeb CA, Saunders NC. 1987. Intraspecific 1657 1660. phylogeography: the mitochondrial DNA bridge between Cognato AI, Seybold SJ, Sperling FAH. 1999. Incomplete population genetics and systematics. Annual Review of barriers to mitochondrial gene flow between pheromone Ecology and Systematics 18: 489–522. races of the North American pine engraver, Ips pini (Say). Avtzis DN, Arthofer W, Stauffer C. 2008. Sympatric occur- Proceedings of the Royal Society of London Series B, Biolog- – rence of diverged mtDNA lineages of Pityogenes chalcogra- ical Sciences 266: 1843 1850. phus (Coleoptera, Scolytinae) in Europe. Biological Journal Cognato AI, Harlin AD, Fisher ML. 2003. Genetic struc- of the Linnaean Society 94: 331–340. ture among pinyon pine beetle populations (Scolytinae: Ips – Behura SK. 2006. Molecular marker systems in insects: cur- confusus). Environmental Entomology 30: 1262 1270. € € rent trends and future avenues. Molecular Ecology 15: Comps B, Gomory D, Letouzey J, Thiebaut B, Petit RJ. 3087–3113. 2001. Diverging trends between heterozygosity and allelic Bohonak AJ. 2002. IBD (Isolation By Distance): a program richness during postglacial colonization in the European – for analyses of isolation by distance. Journal of Heredity beech. Genetics 157: 389 397. 93: 153–154. Cox K, Thomaes A, Antonini G, Zilioli M, De Gelas K, Bosso L, Rebelo H, Garonna AP, Russo D. 2013. Model- Harvey D, Solano E, Audisio P, McKeown N, Shaw P, ling geographic distribution and detecting conservation Minetti R, Bartolozzi L, Mergeay J. 2013. Testing the gaps in Italy for the threatened beetle Rosalia alpina. Jour- performance of a fragment of the COI gene to identify west- nal for Nature Conservation 21: 72–80. ern Palaearctic stag beetle species (Coleoptera, Lucanidae). – Brus R. 2010. Growing evidence for the existence of glacial ZooKeys 365: 105 126. refugia of European beech (Fagus sylvatica L.) in the Dinca V, Runquist M, Nilsson M, Vila R. 2013. Dispersal, south-eastern Alps and north-western Dinaric Alps. Period- fragmentation and isolation shape the phylogeography of icum Biologorum 112: 239–246. the European lineages of Polyommatus (Agrodiaetus) ripar- Buse J, Schroder B, Assmann T. 2007. Modelling habitat tii (Lepidoptera: Lycaenidae). Biological Journal of the Lin- – and spatial distribution of an endangered longhorn nean Society 109: 817 829.

© 2015 The Linnean Society of London, Biological Journal of the Linnean Society, 2015, , – POPULATION GENETIC STRUCTURE OF ROSALIA ALPINA 13

Drag L, Cizek L. 2014. Successful reintroduction of an Gratton P, Konopinski MK, Sbordoni V. 2008. Pleis- endangered veteran tree specialist: conservation and genet- tocene evolutionary history of the Clouded Apollo (Parnas- ics of the great capricorn beetle (Cerambyx cerdo). Conser- sius mnemosyne): genetic signatures of climate cycles and a vation Genetics 16: 267–276. ‘time-dependent’ mitochondrial substitution rate. Molecular Drag L, Hauck D, Pokluda P, Zimmermann K, Cizek L. Ecology 17: 4248–4262. 2011. Demography and dispersal ability of a threatened Grove SJ. 2002. Saproxylic insect ecology and the sustain- saproxylic beetle: a mark-recapture study of the rosalia able management of forests. Annual Review of Ecology and longicorn (Rosalia alpina). PLoS ONE 6: e21345. Systematics 33: 1–23. Drag L, Zima J Jr, Cizek L. 2013. Characterization of nine Haran J, Koutroumpa F, Magnoux E, Roques A, Roux polymorphic microsatellite loci for a threatened saproxylic G. 2015. Ghost mtDNA haplotypes generated by fortuitous beetle Rosalia alpina (Coleoptera: Cerambycidae). Conser- NUMTs can deeply disturb infra-specific genetic diversity vation Genetics Resources 5: 903–905. and phylogeographic pattern. Journal of Zoological System- Duelli P, Wermelinger B. 2005. Der Alpenbock (Rosalia atics and Evolutionary Research 53: 109–115. alpina). Ein seltener Bockkafer€ als Flaggschiff-Art. Merk- Hedl R, Kopecky M, Komarek J. 2010. Half a century of blatt fur€ die Praxis 39: 1–8. succession in a tem-perate oakwood: from species-rich Earl DA, vonHoldt BM. 2012. Structure Harvester: a web- community to mesic forest. Diversity and Distributions 16: site and program for visualizing Structure output and 267–276. implementing the Evanno method. Conservation Genetetics Hewitt GM. 1999. Post-glacial recolonization of European Resources 4: 359–361. Biota. Biological Journal of the Linnean Society 68: 87–112. Ersts PJ 2015. Geographic Distance Matrix Generator, Heyrovsky L. 1955. Fauna CSR. Svazek 5. Tesarıkovitı. Version 1.2.3. Available at: http://biodiversityinformatics. [Fauna of the Czechoslovak Republic. Volume 5. Longhorn amnh.org/open_source/gdmg. Beetles.]. Prague, CSAV. Evanno G, Regnaut S, Goudet J. 2005. Detecting the Horak J, Chumanova E, Hilszczanski J. 2012. Saproxylic number of clusters of individuals using the software Struc- beetle thrives on the openness in management: a case study ture: a simulation study. Molecular Ecology 14: 2611–2620. on the ecological requirements of Cucujus cinnaberinus Excoffier L, Lischer HEL. 2010. Arlequin suite ver 3.5: a from Central Europe. Insect Conservation and Diversity 5: new series of programs to perform population genetics anal- 403–413. yses under Linux and Windows. Molecular Ecology Horn A, Roux-Morabito G, Lieutier F, KerdelhueC. Resources 10: 564–567. 2006. Phylogeographic structure and past history of the cir- Fahrig L. 2013. Rethinking patch size and isolation effects: cum-Mediterranean species Tomicus destruens Woll. the habitat amount hypothesis. Journal of Biogeography (Coleoptera: Scolytinae). Molecular Ecology 15: 1603–1615. 40: 1649–1663. Horn A, Stauffer C, Lieutier F, Kerdelhue C. 2009. Com- Feder JL, Opp SB, Wlazlo B, Reynolds K, Go W, Spisak plex postglacial history of the temperate bark beetle Tomi- S. 1994. Host fidelity is an effective premating barrier cus piniperda L. (Coleoptera, Scolytinae). Heredity 103: between sympatric races of the apple maggot fly. Proceed- 238–247. ings of the National Academy of Sciences of the United Hovorka W. 2011. Nachweise des Alpenbocks (Rosalia States of America 91: 7990–7994. alpina, L.) bei Rabensburg im Weinviertel (Coleoptera: Cer- Ferrari J, West JA, Via S, Godfray HCJ. 2012. Population ambycidae). Beitrage€ zur Entomofaunistik 12 – Kurzmit- € genetic structure and secondary symbionts in host-associ- teilungen. Wien: Osterreichische Gesellschaft fur€ ated populations of the pea aphid complex. Evolution 66: Entomofaunistik. 375–390. Jakobsson M, Rosenberg NA. 2007. CLUMPP: a cluster Frankham R, Ballou JD, Briscoe DA. 2002. Introduction matching and permutation program for dealing with label to conservation genetics. Cambridge: Cambridge University switching and multimodality in analysis of population Press. structure. Bioinformatics 23: 1801–1806. Garbarino M, Sibona E, Lingua E, Motta R. 2014. Decline Jendek B, Jendek E. 2006. An analysis of the beetle con- of traditional landscape in a protected area of the southwest- servation in the Slovakia based on the longicorn beetles ern Alps: the fate of enclosed pasture patches in the land (Coleoptera, Cerambycidae) as a model group. Folia faunis- mosaic shift. Journal of Mountain Science 11: 544–554. tica Slovaca 11: 15–28. Gassert F, Schulte U, Husemann M, Ulrich W, Rodder€ Jensen JL, Bohonak AJ, Kelley ST. 2005. Isolation by dis- D, Hochkirch A, Engel E, Meyer J, Habel JC. 2013. tance, web service, Version 3.23. Available at: http://ib- From southern refugia to the northern range margin: dws.sdsu.edu/ genetic population structure of the common wall lizard, Po- Kadlec T, Vrba P, Kepka P, Schmitt T, Konvicka M. 2010. darcis muralis. Journal of Biogeography 40: 1475–1489. Tracking the decline of once-common butterfly: delayed ovipo- Gepp J. 2002. Rosalia alpina L. – Osterreichs€ Insekt des sition, demography and population genetics in the Hermit, Jahres 2001. Entomologica Austriaca 5: 3–4. Chazara briseis. Animal Conservation 13: 172–183. Goudet J 2002. Fstat, a Program to estimate and test gene Librado P, Rozas J. 2009. DnaSP v5: a software for com- diversities and fixation indices, Version 2.9.3.2. Available prehensive analysis of DNA polymorphism data. Bioinfor- at: http://www.unil.ch/popgen/softwares/fstat.htm matics 25: 1451–1452.

© 2015 The Linnean Society of London, Biological Journal of the Linnean Society, 2015, , – 14 L. DRAG ET AL.

Lindhe A, Jeppsson T, Ehnstrom€ B. 2011. Longhorn bee- may compensate for lower amounts of dead wood in driving tles in Sweden – changes in distribution and abundance richness of saproxylic beetles. Ecography 38: 499–509. over the last two hundred years. Sveriges Entomologiska Nei M, Chesser RK. 1983. Estimation of fixation indices Forening:€ Farjestaden.€ and gene diversities. Annals of Human Genetics 47: 253– Lis A, Maryanska-Nadachowska A, Lachowska-Cierlik 259. D, Kajtoch L. 2014. The secondary contact zone of phylo- Oleksa A, Chybicki IJ, Gawronski R, Svensson GP, genetic lineages of the Philaenus spumarius (Hemiptera: Burczyk J. 2013. Isolation by distance in saproxylic bee- Aphrophoridae: Cercopidae): an example of incomplete allo- tles may increase with niche specialization. Journal of patric speciation. Journal of Insect Science 14. doi:10.1093/ Insect Conservation 17: 219–233. jisesa/ieu089. Oleksa A, Chybicki IJ, Larsson MC, Svensson GP, Magri D. 2008. Patterns of post-glacial spread and the Gawronski R. 2015. Rural avenues as dispersal corridors extent of glacial refugia of European beech (Fagus sylvat- for the vulnerable saproxylic beetle Elater ferrugineus in a ica). Journal of Biogeography 35: 450–463. fragmented agricultural landscape. Journal of Insect Con- Magri D, Vendramin GG, Comps B, Dupanloup I, Gebu- servation 19: 567–580. rek T, Gom€ ory€ D, Latalowa M, Litt T, Paule L, Roure Peakall R, Smouse PE. 2006. GENALEX 6: genetic analy- JM, Tantau I, Knaap WO, Petit RJ, Beaulieu JL. 2006. sis in Excel. Population genetic software for teaching and A new scenario for the Quaternary history of European research. Molecular Ecology Notes 6: 288–295. beech populations: paleobotanical evidence and genetic con- Peakall R, Smouse PE. 2012. GenAlEx 6.5: genetic analysis sequences. New Phytologist 171: 199–221. in Excel. Population genetic software for teaching and Mantel N. 1967. Detection of disease clustering and a gener- research-an update. Bioinformatics 28: 2537–2539. alized regression approach. Cancer Research 27: 209–220. Picard F. 1929. Faune de France. Paris: Coleopteres. Ceram- Marigo G, Peltier J-P, Girel J, Pautou G. 2000. Success bycidae. in the demographic expansion of Fraxinus excelsior L. Trees Pritchard JK, Stephens M, Donnelly P. 2000. Inference 15: 1–13. of population structure using multilocus genotype data. Michalcewicz J, Ciach M. 2012. Rosalia longicorn Rosalia Genetics 155: 945–959. alpina (L.) (Coleoptera: Cerambycidae) uses roadside Euro- Ranius T. 2002. Osmoderma eremita as an indicator of spe- pean ash trees Fraxinus excelsior L. – an unexpected habi- cies richness of beetles in tree hollows. Biodiversity and tat of an endangered species. Polish Journal of Entomology Conservation 11: 931–941. 81: 49–56. Raymond M, Rousset F. 1995. GENEPOP (version 1.2): Michalcewicz J, Ciach M. 2015. Current distribution of population genetics software for exact tests and ecumeni- the Rosalia longicorn Rosalia alpina (LINNAEUS, 1758) cism. Journal of Heredity 86: 248–249. (Coleoptera: Cerambycidae) in Poland. Polish Journal of Rosenberg NA. 2004. DISTRUCT: a program for the graphi- Entomology 84: 9–20. cal display of population structure. Molecular Ecology Notes Michalcewicz J, Bodziarczyk J, Ciach M. 2013. Develop- 4: 137–138. ment of the rosalia longicorn Rosalia alpina (L.) (Coleop- Rousset F. 2008. Genepop’007: a complete reimplementation tera: Cerambycidae) in the sycamore maple Acer of the Genepop software for Windows and Linux. Molecular pseudoplatanus L. – the first report from Poland. Polish Ecology Resources 8: 103–106. Journal of Entomology 82: 19–24. Russo D, Cistrone L, Garonna A. 2011. Habitat selection Miklın J, Cızek L. 2014. Erasing a European biodiversity by the highly endangered long-horned beetle Rosalia hot-spot: open woodlands, veterantrees and mature forests alpina in Southern Europe: a multiple spatial scale assess- succumb to forestry intensification, succession, and logging ment. Journal of Insect Conservation 15: 685–693. in a UNESCO Biosphere Reserve. Journal for Nature Con- Salle A, Artofer W, Lieutier F, Stauffer C, KerdelhueC. servation 22: 35–41. 2007. Phylogeography of a host-specific insect: genetic Miller MP, Haig SM, Wagner RS. 2005. Conflicting pat- structure of Ips typographus in Europe does not reflect the terns of genetic structure produced by nuclear and mito- past fragmentation of its host. Biological Journal of the chondrial markers in the Oregon slender salamander Linnean Society 90: 239–246. (Batrachoseps wrighti): implications for conservation efforts Sama G 2002. Atlas of Cerambycidae of Europe and Mediter- and species management. Conservation Genetics 6: 275– anean Area. Vol. I. Nothern, Western, Central and Eastern 287. Europe, Britisch Isles and Continental Europe from France Morin PA, Luikart G, Wayne RK. 2004. SNPs in ecology, (excl.Corsica) to Scandinavia and Urals.Zlın, Kabourek. evolution and conservation. Trends in Ecology & Evolution Schmitt T. 2007. Molecular Biogeography of Europe: pleis- 19: 208–216. tocene cycles and postglacial trends. Frontiers in Zoology 4: Moritz C. 1994. Defining ‘Evolutionarily significant units’ 11. for conservation. Tree 9: 373–375. Schmitt T. 2009. Biogeographical and evolutionary impor- Muller€ J, Brustel H, Brin A, Bussler H, Bouget C, Ober- tance of the European high mountain systems. Frontiers in maier E, Heidinger IMM, Lachat T, Forster€ B, Horak J, Zoology 6: 9. Prochazka J, Kohler€ F, Larrieu L, Bense U, Isacsson G, Sebek P, Altman J, Platek M, Cizek L. 2013. Is active Zapponi L, Gossner MM. 2015. Increasing temperature management the key to the conservation of saproxylic

© 2015 The Linnean Society of London, Biological Journal of the Linnean Society, 2015, , – POPULATION GENETIC STRUCTURE OF ROSALIA ALPINA 15

biodiversity? Pollarding promotes the formation of tree hol- Tamura K, Stecher G, Peterson D, Filipski A, Kumar S. lows. PLoS ONE 8: e60456. 2013. MEGA6: molecular Evolutionary Genetics Analysis Seibold S, Brandl R, Buse J, Hothorn T, Schmidl J, Version 6.0. Molecular Biology and Evolution 30: Thorn S, Muller€ J. 2014. Association of extinction risk of 2725–2729. saproxylic beetles with ecological degradation of forests in Templeton AR, Crandall KA, Sing CF. 1992. A cladistic Europe. Conservation Biology 1–9. analysis of phenotypic associations with haplotypes inferred Serafim R, Maincan S. 2008. Data on Cerambycidae and from restriction endonuclease mapping and DNA sequence Chrysomelidae (Coleoptera: Chrysomeloidea) from Bucur- data III. Cladogram estimation. Genetics 132: 619–633. esti and surroundings. Travaux du Museum National d’His- Theissinger K, Balint M, Feldheim KA, Haase P, Johan- toire Naturelle 51: 387–416. nesen J, Laube I, Pauls SU. 2013. Glacial survival and Simon C, Frati F, Beckenbach A, Crespi B, Liu H, post-glacial recolonization of an arctic–alpine freshwater Flook P. 1994. Evolution, weighting and phylogenetic util- insect (Arcynopteryx dichroa, Plecoptera, Perlodidae) in ity of mitochondrial gene sequences and a compilation of Europe. Journal of Biogeography 40: 236–248. conserved PCR primers. Annals of the Entomological Soci- Tison J-L, Edmark VN, Sandoval-Castellanos E, Van ety of America 87: 651–701. Dyck H, Tammaru T, Valim€ aki€ P, Dalen L, Gotthard Slama MEF. 1998. Tesarıkovitı – Cerambycidae Ceske repub- K. 2014. Signature of post-glacial expansion and genetic liky a Slovenske republiky (Brouci – Coleoptera). [Ceramby- structure at the northern range margin of the speckled cidae of the Czech Republic and the Slovak Republic]. wood butterfly. Biological Journal of the Linnean Society Krhanice, Milan Slama. 113: 136–148. Slatkin M. 1987. Gene flow and the geographic structure of Vandewoestijne S, Van Dyck H. 2010. Population genetic natural populations. Science 236: 787–792. differences along a latitudinal cline between original and Solano E, Mancini E, Ciucci P, Mason F, Audisio P, An- recently colonized habitat in a butterfly. PLoS ONE 5: tonini G. 2013. The EU protected taxon Morimus funereus e13810. Mulsant, 1862 (Coleoptera: Cerambycidae) and its western Via S, Bouck AC, Skillman S. 2000. Reproductive isolation Palaearctic allies: systematics and conservation outcomes. between divergent races of pea aphids on two hosts. II. Conservation Genetics 14: 683–694. Selection against migrants and hybrids in the parental Song H, Buhay JE, Whiting MF, Crandall KA. 2008. environment. Evolution 54: 1626–1637. Many species in one: DNA barcoding overestimates the Wielstra BM, Crnobrnja-Isailovic J, Litvinchuk SN, number of species when nuclear mitochondrial pseudogenes Reijnen BT, Skidmore AK, Sotiropoulos K, Toxopeus are coamplified. Proceedings of the National Academy of AG, Tzankov N, Vukov TD, Arntzen JW. 2013. Tracing Sciences of the United States of America 105: 13486–13491. glacial refugia fo triturus newts based on mitochondrial Stireman JO, Nason JD, Heard SB. 2005. Hostassociated DNA phylogeography and species distribution modeling. genetic differentiation in phytophagous insects: general Frontiers in Zoology 10: 13. phenomenon or isolated exceptions? Evidence from a gold- Williams BL, Brawn JD, Paige KN. 2003. Landscape scale enrod-insect community. Evolution 59: 2573–2587. genetic effects of habitat fragmentation on a high gene flow Sunnucks P. 2000. Efficient genetic markers for population species: Speyeria idalia (Nymphalidae). Molecular Ecology biology. Tree 15: 199–203. 12: 11–20. Takezaki N, Nei M, Tamura K. 2010. POPTREE2: Software Zakharov EV, Hellmann JJ. 2008. Genetic differentiation for constructing population trees from allele frequency data across a latitudinal gradient in two co-occurring butterfly and computing other population statistics with Windows-in- species: revealing population differences in a context of cli- terface. Molecular Biology and Evolution 27: 747–752. mate change. Molecular Ecology 17: 189–208.

SUPPORTING INFORMATION Additional Supporting Information may be found in the online version of this article at the publisher’s web-site: Figure S1. Barplots gained from STRUCTURE (K = 2, K = 3, K = 4, K = 5, K = 6), and the number of clusters described as the highest number of DK (Evanno et al., 2005) and the mean log probability of data LnP(D). We used 100 000 burn-in and 1 000 000 Markov chain Monte Carlo steps for 10 replicates for each K. Table S1. Null allele frequencies for each locus and population estimated by FREENA.

Table S2. Pairwise FST values for 32 populatins created in ARLEQUIN with 10 000 permutations. The FST values with P > 0.05 are shown in red.

© 2015 The Linnean Society of London, Biological Journal of the Linnean Society, 2015, , –